On 1/29/2013 3:43 PM, usher99@xxxxxxx wrote:
> Trying to eek out one more horse trip (somewhere near Moab I hear
> fromt the CVO) with current horse cam but likely will absolutely
> require frequent HDR
> processing than I am accustomed to due to limited DR. (this could be a
> mistake but running out of time) I did some trial runs with PS with 3
> raw files hand held and is a bunch of work not have them look "off" and
> overdone but results can be fine. The alignment algorithm does seem up
> to the job (I hope).
G9, right? Same sensor as the A650, but possibly better Raw files than the CHDK
ones I got. Bright sun, shoot at ISO 80
and bracket downward from -1/3 to -1 2/3, by 1/3 stops, if possible. That'll
give you plenty for HDR.
But I'm betting you'll almost always find one that holds sufficient highlights
and has recoverable shadows. You can go a
little crazy with NR in the shadows using masking, and later adjust mask(s) and
opacity to lose no visible detail while
knocking down the noise.
I've been amazed how often I bracket, thinking I'll have to combine, and find a
single shot that does the job.
> The lens has a fair amount of barrel distortion at wide end
> and I correct most of it via lens correction module in PS but dial back
> due to some unwanted effects sometimes in the corners as it is not a
> lossless procedure
This is the aspect of the practice of in-camera linear distortion correction
that nobody talks about. I too have seen
considerable loss of detail in the corners or distortion corrected images. I
recall one example that was quite
startling. A clearly recognizable object became a blob
It's inevitable, if you think about how it must be done. For each pixel, a new
location is calculated. What if one pixel
has a calculated movement of 1.5 and its neighbor is 1.4. You can't move by a
partial pixel, so one goes 1 and the
other goes 2. Oops, that straight line now has a kink in it. Multiply that many
times, and you have random movement at
the pixel level.
Because it is a quantization problem, there is no simple way to make it better.
Nothing to do with the quality of the
profile. Because it is strongest in the corners, I occasionally wonder when I
see tests of WA lenses or WA ends of zooms
how much of the corner softness is lens and how much firmware distortion
correction.
In addition to greater computing power in cameras, I suspect part of the reason
makers have gone to this solution to a
lot of other problems in lens design is the high MP counts of contemporary
sensors. The VAST majority of images taken
with them will be displayed at a size where the effect is invisible.
One way to try a better correction might be to upsize the image, do all the
calculations and pixel relocations, then
downsample again. Downsampling would then make combined pixels with in-between
values. Rather than a complex upsample,
simple pixel multiplication might work better. You'd have to test.
Might work. But not in-camera with today's processing power/speed. :-)
> especially for this ye olde cam. If I do that and
> leave the 16bit files open after processing in ACR open in PS Iam
> warned that I will lose some dynamic range (DR) if I don't use the raw
> files. Curiously I did try that and the lens aberrations were clearly
> used by PS in merge to HDR even if not run through ACR first (at least
> by me)--though can not dial back. I am very confused as to what is best
> and how PS knew I wanted to use the lens aberration correction in the
> HDR version when just importing the raw files. Is that adjustable too?
There are two lens aberration correction functions. One in PS itself and one in
ACR. They appear very similar, but are
different in ways the details of which I forget. In any case, there are more
profiles available for the ACR one.
Clearly, the image combine functions use ACR (how else?) to open Raw files.
There are two possibilities here. If you
used the Lens Correction function in ACR, auto or manually adjusted, that is
remembered*, and would be used when the Raw
files are opened. If so, then any manual adjustments you have made to the ACR
Auto settings should carry through.
If you used the PS correction function, and not the one in ACR, it appears that
the Combine default is to turn on the
ACR Auto Correct, whether the file had ever been opened or not and whether you
had used it or not.
Clearly that's they way to do it for panorama stitching, but less clear for
HDR. Perhaps there is a way to change the
setting?
> Puzzled, worried Mike
Follow Meher Baba's ageless wisdom. Don't worry, ride horsey, be happy.
Smilin' Moose
* By default, in 'sidecar' .xmp files, although there is a database option.
--
What if the Hokey Pokey *IS* what it's all about?
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|